Re: [HACKERS] Question about Restart point and checkpoint_segments

[Benoit Lobréau]

Sorry for the noise, I got my answer.

2016-02-16 18:02 GMT+01:00 Benoit Lobréau :

> Hello,
>
> I am using a hot_standby setup on PostgreSQL 9.1
> While I was testing, I found out that only checkpoint_timeout (+ a
> checkpoint since the last restart point) could trigger a restart point.
>
> The code (bgwriter.c) seems to confirm this:
>
> /*
> * Check progress against WAL segments written and checkpoint_segments.
> *
> * We compare the current WAL insert location against the location
> * computed before calling CreateCheckPoint. The code in XLogInsert that
> * actually triggers a checkpoint when checkpoint_segments is exceeded
> * compares against RedoRecptr, so this is not completely accurate.
> * However, it's good enough for our purposes, we're only calculating an
> * estimate anyway.
> */
> if (!RecoveryInProgress())  ===> Only in case of primary
> {
>recptr = GetInsertRecPtr();
>elapsed_xlogs =
>(((double) (int32) (recptr.xlogid -
> ckpt_start_recptr.xlogid)) * XLogSegsPerFile +
>((double) recptr.xrecoff - (double)
> ckpt_start_recptr.xrecoff) / XLogSegSize) /
>CheckPointSegments;
>
>if (progress < elapsed_xlogs)  ===> progress in volume
>{
>ckpt_cached_elapsed = elapsed_xlogs;
>return false;
>}
> }
>
> /*
> * Check progress against time elapsed and checkpoint_timeout.
> */
> gettimeofday(, NULL);
> elapsed_time = ((double) ((pg_time_t) now.tv_sec - ckpt_start_time) +
>
> now.tv_usec / 100.0) / CheckPointTimeout;
>
>
> if (progress < elapsed_time)  ===> progress in time
> {
>ckpt_cached_elapsed = elapsed_time;
>return false;
> }
>
> /* It looks like we're on schedule. */
> return true;
>
> I also found a post from Simon Riggs [1]: "checkpoint_segments is ignored
> on standby."
>
> The documentation is stating the opposite [2]: "In standby mode, a
> restartpoint is also triggered if checkpoint_segments log segments have
> been replayed since last restartpoint and at least one checkpoint record
> has been replayed."
>
> Since I am not a native english speaker, maybe I misunderstood the
> documentation. But to me, it looks wrong.
> If it's indeed wrong. Could you explain why checkpoint_segments doesn't
> trigger a restart_point in standby mode ?
>
> Thank you
> Benoit
>
> [1]
> http://www.postgresql.org/message-id/CA+U5nMKdf7odZzYNnoRkkCZmJpGEy=oqbu9nan_zva_rtzi...@mail.gmail.com
> [2] http://www.postgresql.org/docs/9.1/static/wal-configuration.html
>

[HACKERS] Question about Restart point and checkpoint_segments

[Benoit Lobréau]

Hello,

I am using a hot_standby setup on PostgreSQL 9.1
While I was testing, I found out that only checkpoint_timeout (+ a
checkpoint since the last restart point) could trigger a restart point.

The code (bgwriter.c) seems to confirm this:

/*
* Check progress against WAL segments written and checkpoint_segments.
*
* We compare the current WAL insert location against the location
* computed before calling CreateCheckPoint. The code in XLogInsert that
* actually triggers a checkpoint when checkpoint_segments is exceeded
* compares against RedoRecptr, so this is not completely accurate.
* However, it's good enough for our purposes, we're only calculating an
* estimate anyway.
*/
if (!RecoveryInProgress())  ===> Only in case of primary
{
   recptr = GetInsertRecPtr();
   elapsed_xlogs =
   (((double) (int32) (recptr.xlogid -
ckpt_start_recptr.xlogid)) * XLogSegsPerFile +
   ((double) recptr.xrecoff - (double)
ckpt_start_recptr.xrecoff) / XLogSegSize) /
   CheckPointSegments;

   if (progress < elapsed_xlogs)  ===> progress in volume
   {
   ckpt_cached_elapsed = elapsed_xlogs;
   return false;
   }
}

/*
* Check progress against time elapsed and checkpoint_timeout.
*/
gettimeofday(, NULL);
elapsed_time = ((double) ((pg_time_t) now.tv_sec - ckpt_start_time) +

now.tv_usec / 100.0) / CheckPointTimeout;


if (progress < elapsed_time)  ===> progress in time
{
   ckpt_cached_elapsed = elapsed_time;
   return false;
}

/* It looks like we're on schedule. */
return true;

I also found a post from Simon Riggs [1]: "checkpoint_segments is ignored
on standby."

The documentation is stating the opposite [2]: "In standby mode, a
restartpoint is also triggered if checkpoint_segments log segments have
been replayed since last restartpoint and at least one checkpoint record
has been replayed."

Since I am not a native english speaker, maybe I misunderstood the
documentation. But to me, it looks wrong.
If it's indeed wrong. Could you explain why checkpoint_segments doesn't
trigger a restart_point in standby mode ?

Thank you
Benoit

[1]
http://www.postgresql.org/message-id/CA+U5nMKdf7odZzYNnoRkkCZmJpGEy=oqbu9nan_zva_rtzi...@mail.gmail.com
[2] http://www.postgresql.org/docs/9.1/static/wal-configuration.html